Académique Documents
Professionnel Documents
Culture Documents
1, February 2013
Manuscript received August 14, 2012; revised November 9, 2012. A. Description of the Standard Release Model (SRM)
The authors are with the Process & Quality Consulting team with Traditionally, the IT division of the bank was responsible
Cognizant Technology Solutions, India. (e-mail: Sujoy.de@ cognizant.com,
amit.rahut@cognizant.com, uttam.bhattacharya@cognizant.com). for maintaining their legacy mainframe applications. They
103
Lecture Notes on Software Engineering, Vol. 1, No. 1, February 2013
B. Analysis of the Branch for Testing Model The resulting table is displayed in Table I.
This model successfully addressed the various drawbacks As expected, month four saw a huge spike in SRM as most
in the Standard Release Model as follows: of the bug fixes were naturally done in month four (last
1) As development and product documentation could go on month before deployment). In BTM, the merging effort
uninterrupted in the mainline (trunk), Requirements predictably diminished in the last month as the code base was
Engineers and Solution Architects could move on to the fairly stable when the branch was created and most of the
documentation for the next release immediately after fixes were already completed by month three. Comparison of
freezing them for the current release. total effort demonstrates that BTM is a clear winner
2) The team could open release branches very late into the compared to SRM.
release. At this time, the code development of the RE, SE Idle Time: Last but not the least, the idle time in
complete functionality of the current release would have person days for the team was also recorded. The results are
been completed and only testing and bug fixes would tabulated in Table II.
remain before it was deployed. Thus the code-base in the By eliminating the Requirements Engineers’ and Solution
mainline would be in a very stable state when branched. Engineers’ waiting time for a new branch creation, BTM
3) Since only fixes for the bugs found in System Testing ended up with huge effort (and hence cost) savings over SRM.
were done in the release branches and no development Here also, BTM came out superior to SRM.
took place there, the number of files that required The experiment with the JAP prototype showed a total
merging was much lesser, resulting in much lesser effort savings of 136 + (22 – 3) / 8 ≈ 138 person days in the
merging effort in total. BTM model when compared with SRM. Although the data
4) This was also in line with the golden CM principle of was captured using a prototype, the results prove
branch late, merge early. conclusively that Branch for Testing Model is more efficient
However, this model was also not completely foolproof. than the Standard Release Model.
Some of its disadvantages were: TABLE I: TOTAL MERGING EFFORT COMPARISON
1) The approach is not intuitively easy to understand for SRM BTM
someone uninitiated to CM principles. Baselines for a Month 1 0 0
release have to be taken partly on the mainline (trunk)
Month 2 0 0
and partly on the release branch, which can become
Month 3 6 2
confusing.
2) A conflicting requirement over two simultaneous Month 4 16 1
releases is a bit of a challenge and requires careful Total 22 3
handling. E.g., functionality in Release 3.0.0 that
TABLE II: RE, SE IDLE TIME COMPARISON
conflicts with functionality for Release 2.0.0 will need
some special handling from the project team, as SRM BTM
development for release 3.0.0 being done on the trunk Month 1 0 0
will normally be overridden by the regular merges to it Month 2 80 0
from the Release 2.0.0 branch till the time Release 2.0.0 Month 3 56 0
is deployed into production. Month 4 0 0
Total 136 0
104
Lecture Notes on Software Engineering, Vol. 1, No. 1, February 2013
of the organization on the other. framework, metrics definition for a client organization. He has worked as a
Configuration Manager in the IT division of the largest private bank in
Europe. He has experience in organization wide implementation of process
REFERENCES management applications for application development and maintenance
[1] M. Sakyo. (August 2004). The challenges posed by parallel projects and has an in-depth understanding of SDLC concepts, continual
development on release management. CM Crossroads™ [Online]. improvements and high maturity process areas. He has worked as a Quality
Available: Lead for process benchmarking and implementation for a big manufacturing
http://www.cmcrossroads.com/cm-journal-articles/6740-the-challenge organization and had implemented Theory of Constraint project resulting in
s-posed-by-parallel-development-on-release-management increased profitability. Mr. Rahut is certified Project Management
[2] B. C. Sussman, B. W. Fitzpatrick, and C. M. Pilato, Version Control Professional (PMP®) from PMI, USA and has cleared the ITIL® version 3
with Subversion, For Subversion 1.7: (Compiled from r4304), pp. 96. Foundation Examination from EXIN/APMG/OGC. He is also a certified
[3] CMMI for Development, Version 1.2, CMMI-DEV, V1.2, Carnegie Software Quality Analyst from Cognizant Certified Professional
Mellon, Software Engineering Institute, 2006, pp. 114-130. examination. Mr. Rahut was also associated with American Quality Society
[4] L. Wingerd, Practical Perforce, 1st ed. O’Reilly Media, 2005, vol. 7, (ASQ) for two years and was a member of Project Management Institute
pp. 178-180. (PMI), USA. He is also an eminent writer in the Cognizant Process Quality
[5] J. D. Meier, J. Taylor, A. Mackman, P. Bansode, and K. Jones, Team Consulting newsletter and is part of the editorial board.
Development with Visual Studio Team Foundation Server patterns &
practices, Microsoft Corporation, vol. 5.
[6] J. Buffenbarger and K. Gruell, “A branching/merging strategy for Uttam Bhattacharya is a Senior Consulting Manager
parallel software development,” in Proc. of 9th International of Cognizant Technology Solutions having 19 Years of
Symposium, SCM-9, Toulouse, France, September 1999, pp. 1-2. experience in the field of strategic assessment, process
definition, implementation and process improvement in
Sujoy De is a consultant of Cognizant Technology CMMI, Six Sigma, and ISO 9001. Mr. Bhattacharya
Solutions having 8 years of experience in various fields was born in Kolkata, India on 2nd August, 1970 and
of Software Quality and Tool Implementation. Mr. De obtained his engineering graduation (Bachelor in
was born in Bankura, India on 28th of July, 1981 and Technology) in the year 1993 from Calcutta University,
received his engineering degree (Bachelor in Computer India. Mr. Bhattacharya has also completed his MBA
Science & Engineering) in the year 2004 from Burdwan (part time) from Calcutta University, India in 2001. He had played the role of
University, India, and Diploma in Business Quality manager for Cognizant and was responsible for ensuring quality of
Administration in the year 2009 from Pune University, deliverables of the projects. He has implemented CMMI, Six Sigma, ISO
India. He has wide experience in various fields of software quality like 9001 framework, metrics definition for various business units in Cognizant.
Process definition & implementation, process improvement and maintaining He has also led the CMMI assessment for Cognizant. He has wide experience
the Quality Management System. He has also experience in CMMI Level 3 in the field of consulting with direct interfacing with many clients for
implementation, ISO 9001 framework and metrics definition. He has worked Strategic assessment, Process definition, implementation, improvement and
as a Configuration Manager for the IT division of one of the largest private maintaining their Quality Management System for the client organizations
banks in Europe. He has experience in organization wide implementation of spread across geographies. He has also led a number of Six Sigma projects.
process management applications for application development and He has wide experience in organization wide implementation of various
maintenance projects and has an in-depth understanding of SDLC concepts, processes in different types of projects and has an in-depth understanding of
continual improvements and high maturity process areas. In his previous SDLC concepts, continual improvements and high maturity process areas.
organization, he was instrumental in the organization's achieving the ISO Mr. Bhattacharya is a certified Project Management Professional (PMP®)
9001:2000 recertification and its preparation for ISO 140001 certification. from PMI, USA and has cleared the ITIL® version 3 Foundation
Examination from Quint. He is also a certified Six Sigma Black Belt
Certification form BMG, and is a certified internal auditor of ISO 9000. Mr.
Amit Kumar Rahut is a consultant of Cognizant Bhattacharya is a certified Scrum master from Scrum Alliance and is a
Technology Solutions having 10 Years of experience in member of Project Management Institute (PMI), USA. He is also an eminent
the field of process definition, implementation and writer in the Cognizant Process Quality Consulting newsletter and is part of
process improvement with CMMI, Six Sigma, and ISO the editorial board.
9001 model. Mr. Rahut was born in Kolkata, India on
31st October, 1977 and became an engineering graduate
(Bachelor in Technology) in the year 2002 from Calcutta
University, India. He has wide experience in the field of
consulting with direct interfacing for many clients for process definition,
implementation, process improvement and maintaining their Quality
Management System. He has implemented CMMI, Six Sigma, ISO 9001
105